Float-constrained approval-load smoothing: code, synthetic 100-programme ensemble and aggregate data

Description

This record accompanies the manuscript “Schedule mining of engineering approval cycles to reduce planned information congestion in EPC delivery” by Ahmad Baghdadi. It contains Python code for float-constrained approval-load smoothing (FCALS), exact and alternative benchmarks, 13 unit tests, 100 deterministically generated synthetic approval programmes (13,426 artificial cycle records), and 200 run-level outcomes under 28- and 56-day displacement caps. It also includes non-identifying relative-time figure data and published aggregate case tables. The synthetic mean peak reductions are 42.4265% and 48.8319% under the two caps. The code and synthetic inputs reproduce the method and synthetic robustness analysis. The confidential EAP-1 source schedule and real cycle-level records are withheld; this public record cannot independently regenerate the private case extraction or exact case benchmark. No project or company names, exact dates, WBS identifiers, activity codes, or raw precedence edges are included. Code is licensed under MIT; data and documentation under CC BY 4.0, as documented in the archive.

Steps to reproduce

Unpack the ZIP archive. Use Python 3.12 or later and install code/requirements.txt. From the code directory, run: python -m unittest discover -s tests -v; python run_synthetic_demo.py; python run_synthetic_ensemble.py --instances 100 --workers 4 --output rerun_results.csv. Compare rerun_results.csv with data/synthetic_ensemble_results.csv; with the pinned dependencies they are byte-identical. The 100 artificial input programmes can be generated with code/generate_synthetic_demo.py using seed 810000 + index and cycle count 90 + (index * 37) mod 91 for indices 0–99. The README and SHA256SUMS.txt document the files and checksums. The confidential real schedule is not distributed, so the private case extraction cannot be independently rerun.

Institutions

• Umm al-Qura University

  Mecca Region, Makkah

Categories

Licence